Safe Vehicle Distance Monitoring System Based On Monocular Vision

The visual system plays an irreplaceable role in the automatic driving system,and in the visual system,monocular vision has the advantage of low cost compared with multi-ocular vision.For the demand of safe driving,safe vehicle distance monitoring is the focus of research in the field of automatic driving.Therefore,based on the monocular vision,this paper studies the position recognition of the vehicle ahead,distance detection,speed detection and the automatic calibration method,and builds a set of safe vehicle distance monitoring experimental system.The feature of this system is that it only relies on monocular vision to realize safe vehicle distance monitoring without other sensors,which reduces the cost and difficulty of installation and use.It can not only be used as the safe vehicle distance monitoring module in the automatic driving system,but also can be used as the safe vehicle distance auxiliary reminder system for the drivers.The main research contents and main works of this paper are described as follows.1.Design the position recognition method.To monitor the safe vehicle distance,the first task is to determine whether the position of the vehicle ahead is on the forward route of this vehicle.Based on the analysis of the position relationship between the vehicle ahead and the lane,this paper has designed two position recognition methods for the vehicle ahead,which are coordinate method and grouping method.The coordinate method is to judge the position of the vehicle ahead according to the position relationship between the left and right endpoints of the front vehicle’s contour and the reference points of the two lane lines which are on the same horizontal line.The grouping method is to count the coordinate set of the contour of the vehicle ahead and the coordinate set of this lane,and identify whether the vehicle ahead is a monitoring vehicle by judging whether the two sets intersect.2.Design the distance detection method.The first scheme is to segment the image according to the actual distance,calculate the actual distance represented by the unit ordinate in each segment,and calculate the vertical distance between the two vehicles according to the number of unit ordinate between the reference points of the monitoring vehicle and this vehicle.The second scheme is to establish the pixel coordinate and the spatial coordinate with the reference point of the vehicle as the origin of coordinates firstly.Then combining the measured data,establish the conversion model between the two coordinates.Finally,the vertical distance between the two vehicles can be obtained by converting the ordinate of the reference point of the monitoring vehicle in the pixel coordinate into the spatial coordinate.3.Design the speed detection method.The size of safe vehicle distance depends on the current speed of this vehicle,so the speed of this vehicle should be detected in real time.The speed detection method designed in this paper is based on the coordinate conversion model of the vehicle distance detection.The distance which this vehicle has run is calculated by the position change of the reference point of the lane line in two adjacent frames of the image,and then the average speed in the interval time between two frames is obtained,which is approximated to the instantaneous speed.4.Design an automatic calibration method.Since the difference of camera installation position and angle have a great impact on the accuracy of the vehicle distance detection method based on monocular vision,the automatic calibration function is added to meet the requirements of installed in different positions of different vehicles.5.Experimental system test.Select the best effective method to build a safety distance monitoring experimental system and test it.